Personalization for LocationBased ELearning

1
Personalizationfor Location-Based E-Learning
2nd International Conference and Exhibition on
Next Generations Mobile Applications, Services
and Technologies (NGMAST 2008), Cardiff, Wales, UK

• Rui Zhou and Klaus Rechert
• Communication Systems, Dept. of Computer Science
• The University of Freiburg, Germany
• 18.09.2008

2
Outline

• Location-based E-Learning
• Process of personalization
• Personalization techniques
• Learning context modeling
• Content selection and recommendations
• Personalized multimedia presentation
• Prototypical application

3
Location-based E-Learning

• Learners are on the move
• Learners are equipped with portable devices with
  Internet connection
• Keep track of learners
• Provide learners with location-based contents and
  services
• Used in museums, botanical gardens, national
  parks, zoos
• Indoors and outdoors

4
Location-based E-Learning

• Location-based content delivery
• Guiding
• Predefined tours, for beginners
• Tours generated on the fly, based on learners
  requirements and interests
• Recommendations

Location determination
Location-dependent content queries
Compose personalized presentation
Deliver the presentation
5
Location determination

• Primary prerequisite for location-based services
• GPS, Galileo for outdoors
• Wi-Fi fingerprinting for indoors
• 802.11 received signal strength
• Accuracy down to a few meters
• Composite positioning
• Improved accuracy, robustness and multiplicity
• Sensor fusion

6
Process of personalization

• Tailor services
• Facilitate work
• Accommodate social requirements

7
Learning context

• Personalization is based on the context
• Characterizes the situation and environment
• Information about the learner
• Personal profile, goal, knowledge, interests,
  preferences, interaction and presentation
  history
• Information about the environment
• Location, device, time, date, weather

8
Learning context - location

• Location is the most important ingredient
• Contents are provided according to location
• What other learners have learnt at the same
  location will be recommended
• Geographic coordinate or symbolic location
• Orientation, facing objects, velocity, confidence
  about location estimation

9
Learning context

• Personal profile
• Characteristics of the learner
• Name, gender, occupation, nationality
• Initial stereotype -gt initial learning context
  model
• Goal
• To learn a new topic
• To Review an old topic
• An e-course for a conventional lecture
• To prepare an examination
• Contents that other learners with the same goals
  have learnt will be recommended

10
Learning context - knowledge

• Most important for educational systems
• Adapt learning activity to learners knowledge
• Knowledge model is an overlay of domain model
• Domain model
• Expert knowledge of the domain
• A network model
• Decompose to concepts/topics
• Learner knowledge model
• (concept/topic, knowledge level)
• Qualitative or numeric values

From P.Brusilovsky, A.Kobsa, W.Nejdl. The
Adaptive Web. Springer. 2007
11
Learning context - knowledge

• Knowledge update
• Changeable increases (learn) and decreases
  (forget)
• Initially empty
• Updated through interactions
• Learn a topic -gt knowledge increases
• Self test -gt knowledge estimation and update
• Knowledge propagation
• Adaptive learning
• Check knowledge level first
• Present adaptively
• Ranking of recommendations

12
Learning context - interest

• Most commonly used for tourism and learning
• Provide information the learner is interested in
• Interest model is an overlay of domain model
• (concept/topic, interest level)
• Interest update
• Initially empty
• Updated
• Learner chooses a topic -gt interest increases
• Longer learning -gt higher interest level
• Interest propagation

13
Learning context - interest

• Adaptivity
• Check interest level first
• Adaptive presentation
• Low brief information
• Intermediate more detailed information
• high full explanation
• Ranking of recommendations
• Contents that other learners with similar
  interests have learnt will be recommended

14
Learning context

• Preferences
• About how to present
• Multimedia types
• Notification mode upon new presentation
• Presentation mode
• Provided by learners explicitly
• Interaction and presentation history
• Series of presentations cohesive
• Capture the characteristics of the learner

15
Learning context

• Device
• Different portable devices have different OS,
  screen sizes
• Adapt presentation to the device
• Solution
• Map the combinations of features to a few
  stereotypes
• Adapt presentation to the stereotype
• Other context elements
• Time, date, season, weather

16
Content organization

• Local database
• Support location-awareness
• spatial attribute
• point (location) or polygon (region)
• indexed on spatial attribute
• Support personalization
• tagged for stereotypes
• knowledge levels
• interest levels
• others like time and date

17
Location-dependent content query

• New location -gt new learning material
• Planar point query
• Window query
• Nearest neighbor(s) query
• Single query or combination

Planar point query
Window query
Nearest neighbor query
18
Recommendations

• When
• New location
• First location-dependent content queries
• Then recommendations
• After finish learning a topic
• What to learn next
• How
• Based on learning context model
• Collaborative filtering location, goal, interests

19
Ranking of recommendations

• Most relevant topics appear first
• Remove irrelevant topics
• Stereotype
• Knowledge level
• Interest level
• Preferences
• Context elements like time, date and weather
• Rank relevant topics
• Interest level
• Knowledge level
• Distance
• Learnt topics are at the bottom marked learnt

20
Content selection for presentation

• Select content for a topic from local database
• Compose to a personalized multimedia presentation
• Selection is based on
• Stereotype
• Knowledge level
• Interest level
• Context elements such as time, date and weather

21
Personalized multimedia presentation
Modified from A. Scherp. A component framework
for personalized multimedia applications. Ph.D.
dissertation, Dept. Computer Science, University
of Oldenburg, Oldenburg, Germany, 2006.
22
Location-based Botany Guide

• Location-based guiding and learning in botanic
  gardens
• For biology majors and visitors
• Personalized multimedia information of nearby
  plants
• Botanic description of the plant
• Hyperlinks to its botanic parents and ancestors
• Hyperlinks to its botanic children and planted
  individuals
• Recommendations such as what to see next
• WLAN- and GPS-enabled portable device
• GPS for outdoor positioning
• WLAN for indoor positioning and data communication

23
Location-based Botany Guide
24
Browser-Web server architecture
25
Location-based Botany Guide

• Stereotype teacher, biology major, visitor
• Domain model based on botanic taxonomy
• Knowledge model and interest model are overlays
• Qualitative knowledge level
• novice, intermediate, advanced
• determined by scientific self test
• knowledge propagation to parents, to children,
  to siblings
• Qualitative interest level
• low, intermediate, high
• determined by interactions between learner and
  the system
• interest propagation to parents, to children, to
  siblings

26

• Thanks!